A R T I C L E S

Myosin V belongs to the myosin superfamily of actin-based molecularmotors and is involved in the intracellular transport of organelles1–4.Myosin V consists of two identical heavy chains, each composed of anN-terminal motor domain (‘head’), a domain comprising six IQmotifs that bind light chains (‘neck’), a coiled coil dimerizationdomain and a globular cargo-binding tail domain1,3. Myosin V is aprocessive motor that ‘walks’ along an actin filament toward thebarbed end over a long distance without dissociating from the fila-ment5,6. Electron microscopy of actomyosin V in the presence of lowATP concentrations shows both motor domains of myosin V bound tothe actin filament at sites spaced 36 nm apart, which corresponds tothe half pitch of the filament long-pitch helix7. Experiments usingoptical tweezers identified processive 36-nm steps of a bead, on whichsingle myosin V molecules were adsorbed6,8. Moreover, it was shownthat myosin V walks as a left-handed spiral motor along an actin fila-ment, because the average step size is slightly shorter than the halfpitch of the long-pitch actin helix9.

The hand-over-hand walking model has received strong supportfrom two recent experiments that (i) observed the orientation of theneck domain of myosin V by monitoring the polarization of a singlefluorophore covalently attached to a light chain10 and (ii) measuredthe stepwise displacement of a single fluorophore labeled at one of sixlight chains of myosin V11.

Solution kinetic studies demonstrate that ADP release occurs at ∼ 15 s–1 and limits the myosin V ATPase cycle12. Microscopic analysisof myosin V stepping under various nucleotide conditions is consis-tent with rate-limiting ADP release13. The next key target is to deter-mine how the mechanical and biochemical cycles are coupled to eachother at the single-molecule level.

Here, we focused on mechanical events and detected substeps thatoccur within each regular 36-nm step with high temporal resolution.Each regular 36-nm step is composed of two consecutive substeps, onegenerating a 12-nm substep and the other a 24-nm substep. To investi-gate how these substeps and the states attained after the steps are coupled to the ATPase cycle of myosin V, we examined the effects ofATP and ADP concentrations, and 2,3-butanedione 2-monoxime(BDM)14. We also examined the force dependence of the occurrencefrequency of each step and substep, and the dwell time of each state.

RESULTSMovement of myosin V along an actin filamentA single myosin V–coated bead was trapped with optical tweezers andbrought into contact with a fluorescently labeled biotinylated actin filament, which was immobilized on an avidin-coated glass surfacethrough biotinylated BSA (Fig. 1a). A focused red light (685 nm) laserwas used to diagonally illuminate the bead, and its dark-field imagewas projected onto a quadrant photodiode. The bead displacementwas determined by measuring the differential output of the quadrantphotodiode with nanometer accuracy and a 10-kHz sampling rate15.The use of a 200-nm-diameter bead here instead of a 1-µm bead wasessential to obtain a high spatiotemporal resolution.

An example of the time course of bead displacement along an actinfilament (Fig. 1b) shows three consecutive runs of a single myosin Vmolecule along an actin filament at a saturating ATP concentration(1 mM). As the bead began to deviate from the trap center, a positiveexternal load was applied to the myosin V–actin complex (toward thepointed end of an actin filament). Myosin V detached from actin at astall force of ∼ 3 pN. After detachment, the bead quickly returned to the

1Department of Physics, School of Science and Engineering, Waseda University, 3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan. 2Department of Metallurgy,Graduate School of Engineering, 3Center of Interdisciplinary Research, Tohoku University, Sendai, 980-8579, Japan. 4Department of Molecular Biophysics andBiochemistry, Yale University, New Haven, Connecticut 06520, USA. 5Advanced Research Institute for Science and Engineering, Waseda University, 3-4-1 Okubo,Shinjuku-ku, Tokyo 169-8555, Japan. Correspondence should be addressed to S.I. (ishiwata@waseda.jp).

Published online 1 August 2004; doi:10.1038/nsmb806

Mechanochemical coupling of two substeps in a singlemyosin V motorSotaro Uemura1, Hideo Higuchi2,3, Adrian O Olivares4, Enrique M De La Cruz4 & Shin’ichi Ishiwata1,5

Myosin V is a double-headed processive molecular motor that moves along an actin filament by taking 36-nm steps. Using opticaltrapping nanometry with high spatiotemporal resolution, we discovered that there are two possible pathways for the 36-nmsteps, one with 12- and 24-nm substeps, in this order, and the other without substeps. Based on the analyses of effects of ATP,ADP and 2,3-butanedione 2-monoxime (a reagent shown here to slow ADP release from actomyosin V) on the dwell time and theoccurrence frequency of the main and the intermediate states, we propose that the 12-nm substep occurs after ATP binding tothe bound trailing head and the 24-nm substep results from a mechanical step following the isomerization of an actomyosin-ADPstate on the bound leading head. When the isomerization precedes the 12-nm substep, the 36-nm step occurs without substeps.

NATURE STRUCTURAL & MOLECULAR BIOLOGY VOLUME 11 NUMBER 9 SEPTEMBER 2004 877

©20

04 N

atur

e P

ublis

hing

Gro

up

http

://w

ww

.nat

ure.

com

/nat

stru

ctm

olbi

ol

A R T I C L E S

trap center and immediately began to deviate from the trap centeragain as a result of the processive motility of myosin V along actin.

We observed regular forward steps of ∼ 36 nm, approximatelyequal to the half pitch of the actin filament helix. Nearly half of the36-nm steps contained an ‘intermediate state’, indicated by horizon-tal arrows in Fig. 1b (compare inset). At higher forces ≥2 pN, backward 36-nm steps (see large vertical arrows in Fig. 1b) and back-ward steps from the intermediate state (see small vertical arrows inFig. 1b) were identified8,16.

To confirm that a single molecule is sufficient to generate the move-ment observed, we examined the fraction of beads that bind and moveprocessively along an actin filament at various mixing ratios of myosinV and bead. We confirmed by statistical analysis that single myosin Vmolecules were sufficient to move the beads (Fig. 1c)17,18.

The force-velocity relationships in the presence of 1 mM ATP (± ADP) were sigmoidal, showing a steep decrease in the velocity athigher than ∼ 1 pN (Fig. 1d). In comparison, the relationship in thepresence of 10 µM ATP showed a small force dependence. ADP releaseis rate limiting in the presence of 1 mM ATP12, whereas ATP bindingbecomes rate limiting at low ATP concentrations12. The present resultssuggest that ADP release is more load dependent than ATP binding. Itshould be noted that the stall force, 2.5–3 pN, does not depend on thenucleotide conditions (Fig. 1d).

Measurement of substeps within each 36-nm stepThe time course of bead movement examined on an expanded timescale with 0.1-ms time intervals clearly shows the existence of substepswithin the regular 36-nm step (Fig. 2). We identified the presence of

878 VOLUME 11 NUMBER 9 SEPTEMBER 2004 NATURE STRUCTURAL & MOLECULAR BIOLOGY

Figure 1 Stepwise movement of single myosin V motor under variousexternal loads. (a) Schematic illustration showing how the movement of a myosin V–coated bead is measured. To allow the interaction of myosin V with an actin filament, the bead trapped with optical tweezers was movedonto a single actin filament attached to the glass surface through BSA using the biotin-avidin interaction. The position of the trap center was thenfixed. The black and blue arrows, respectively, show the directions for themyosin V movement and the applied external load. (b) An example showingthe displacement of the bead, where the consecutive 36-nm stepwisemovements are clearly seen. A backward 36-nm step is shown by a largevertical arrow. A backward step for the substep is shown by a vertical smallarrow. An intermediate state (shown by horizontal arrows) after a short step(substep) is sometimes observable; the 36-nm step in the middle trace isenlarged in an inset. The force was calculated from the displacement of thebead from the trap center times trap stiffness (0.009 pN nm–1 in b; rightaxis). (c) Relation between proportion of beads that moved along an actinfilament and a mixing molar ratio of myosin V to beads. The proportion ofmoved beads (avg. ± s.d.) was obtained by examining three trials for eachbead (20 different beads) at each point. A solid curve was obtained by fittingwith 1 – e–λC, where c is the mixing molar ratio of myosin V to beads, and λ (0.197) is the fitting parameter9,17,18. (d) Force-velocity relationshipobtained under different conditions (n = 8–24 at each point, total = 405).The relationship shown by solid curves was obtained as described in theMethods section.

Figure 2 The time course of myosin V movement at a 10-kHz sampling rate and forcedependence of the occurrence of various steps.(a) Consecutive two substeps at 1 mM ATPshown in the upper trace and at 1 mM ATP +100 mM BDM shown in the lower trace. Wedetermined whether or not substeps werepresent by obtaining the histogram of the beadpositions at 0.1-ms time intervals (gray lines).Black lines were obtained by the smoothing of21 successive points. (b) Force dependence of step size in the presence of 1 mM ATP. Thesize of the steps was estimated as described inMethods. The first 12-nm substep, blue circle;the second 24-nm substep, green triangle; the36-nm step in which substeps could not beidentified, red square; the 12-nm substep that was followed by a backward step, dark blue circle; the backward steps, black circles. We could notdetermine the occurrence of the 24-nm backward step. The figures along the right ordinate are the average step size estimated independent of force asshown by a straight line parallel to the abscissa.

©20

04 N

atur

e P

ublis

hing

Gro

up

http

://w

ww

.nat

ure.

com

/nat

stru

ctm

olbi

ol

A R T I C L E S

the intermediate state by making a histogram of the bead positionbefore and after each 36-nm step. We refer to each 36-nm step, inwhich substeps could not be identified, as a ‘main step’. The state pop-ulated after the main step or subsequent to the intermediate state istermed a ‘main state’. The intermediate state had a lifetime of a fewmilliseconds at low forces in the presence of 1 mM ATP (an uppertrace in Fig. 2a) and tens of milliseconds in the presence of 100 mMBDM (lower trace in Fig. 2a). The main state had a longer lifetime(several tens of milliseconds) than the intermediate state in all the con-ditions tested. The substeps occurred at random and did not correlatewith the occurrence of the 36-nm main steps.

The step size was determined directly from the time course of step-wise movements of the myosin V–coated bead. A step size distributionhad peaks at 11.6 nm (12-nm substep), 24.1 nm (24-nm substep) and35.4 nm (36-nm step in which substeps could not be identified), andincluded backward steps at 12 nm and 36 nm (Fig. 2b). The step size ofeither the two substeps or for the single main step depended little on theforce level (Fig. 2b), although a possibility that the step size depends onthe force level cannot be excluded because we did not take account of theattenuation factor to estimate the step size (see Methods). On the otherhand, approaching the stall force, the occurrence frequency of 24-nmand 36-nm steps decreased and that of backward steps increased.

Characterization of the intermediate stateHistograms showing the distribution of intermediate state dwell timesunder several nucleotide and force conditions fit single exponentialsregardless of the nucleotide conditions (Fig. 3a,b), indicating that thisdwell time is coupled with a single chemical reaction. The decay time

estimated by exponential fitting became longer as force (balanced toexternal load) increased. It should be noted here that the occurrencefrequency <2 ms dwell time (shown by pink bars in Fig. 3a and thinblack bars in Fig. 3b) largely exceeded that estimated from the expo-nential fitting of the dwell time, strongly suggesting that there is apathway for the 36-nm main step without the intermediate state. Inthe Discussion, we attempt to construct a walking model that canexplain this postulate.

The intermediate dwell time did not depend on the concentrationsof ATP and ADP (Fig. 3c), suggesting that the intermediate state is notcoupled with nucleotide binding or release. On the other hand, BDMmarkedly prolonged the intermediate dwell time (Figs. 2a and 3b).

The force dependence of the average dwell time of the intermediatestate is expressed by the single exponential function of the force, τi exp(Fdi / kBT), where τi is the average dwell time in the absence offorce (F), di the characteristic distance (a parameter having the dimen-sion of length that characterizes the bond instability against appliedload), kB the Boltzmann constant and T the absolute temperature(Fig. 3c). We determined the values of τi and di by global fitting of allthe data19,20, assuming that τi is only prolonged by the addition ofBDM and di is common to all the conditions examined (see Table 1).

Characterization of the main stateIn contrast to the intermediate dwell time (Fig. 3a–c), the histogram ofthe main dwell time (Fig. 3d,e) shows a peak. The main dwell time canbe expressed by the sum of two exponential functions as shown previously8, indicating that the main dwell time is coupled with twoconsecutive chemical reactions.

NATURE STRUCTURAL & MOLECULAR BIOLOGY VOLUME 11 NUMBER 9 SEPTEMBER 2004 879

Figure 3 Force dependence of the intermediate and the main dwell times at various nucleotide states and of theoccurrence frequency of the intermediate state. (a,b) Histogram showing the distribution of the dwell time, τ, ofthe intermediate state in the presence of 1 mM ATP in a and 1 mM ATP + 100 mM BDM in b at different forcelevels. The data <2 ms dwell time (pink bars in a and thin black bars in b) include those in which substeps couldnot be identified. The fitting curves (thin lines) expressed by exp(–τ / τ0) were obtained without the data <2 ms(see text for explanation), where τ0 is a time constant (equal to average dwell time) dependent on developedforce. (c) Force dependence of the average dwell time, τ0, of the intermediate state at 1 mM ATP, 10 µM ATP,1 mM ATP + 200 µM ADP and 1 mM ATP + 100 mM BDM (n = 8–14 at each point, total = 225). The averagedwell time was determined at several force levels different from those shown in a and b. Among the data shown in a and b, only the data in which the substeps were identified were used for the analysis in c. A red line wasdetermined for the data at the first three conditions and a black line for the last condition by using a global fitanalysis assuming the same slope for the two lines. (d,e) Histogram showing the distribution of the main dwelltime, τ, in the presence of 1 mM ATP in d and 1 mM ATP + 100 mM BDM in e at low force level. (f) Force

dependence of the average dwell time, τ0 (= k1–1 + k2

–1) of the main state at 1 mM ATP, 10 µM ATP, 1 mM ATP + 200 µM ADP and 1 mM ATP + 100 mMBDM (n = 18–32 at each point, total = 562). The average dwell time was determined at several force levels different from those shown in d and e. All thedata shown in d and e were used for the analysis in f. The force dependence of τ0 for each condition was expressed by τD exp(FdD / kBT) + τC, where τC isthe time constant representing the dwell time independent of force. The values obtained are summarized in Table 1. (g) The occurrence frequency of theintermediate state at several force levels for each nucleotide condition. Not only the 36-nm steps with substeps but also those without substeps areincluded in the first column shown by pink and thin black bars in a and b. Thus, the occurrence frequency of the 36-nm main steps without substeps wasestimated by subtracting that obtained by extrapolating the exponential function of the dwell time distribution for the intermediate state.

©20

04 N

atur

e P

ublis

hing

Gro

up

http

://w

ww

.nat

ure.

com

/nat

stru

ctm

olbi

ol

A R T I C L E S

Either lowering the ATP concentration or adding ADP caused anincrease in the average dwell time, τ0 (τ0 = k1

–1 + k2–1) (Fig. 3f). This

indicates that ATP binding and ADP release shorten the dwell time ofthe main state8. We also confirmed that the average dwell timeincreases with force (Fig. 3f). We observed that 100 mM BDM pro-longed this dwell time at every force level.

The force (F) dependence of the average main dwell time isexpressed by τD exp(FdD / kBT) + τC, where the values of τD, dD and τCwere determined by global fitting of all the data19,20, assuming that dDis common to all the conditions examined, τD is longer in the presenceof either ADP or BDM, and τC is different at every condition. We con-clude that τD is attributable to ADP release in the absence of load,because the value of τD is determined independent of the ATP concen-tration but prolonged by ADP and BDM (see Table 1). τC is a constantterm, which increased on lowering the ATP concentration and addingADP (see Table 1). Also, it is to be noted that the large value of dD(12.5 nm) indicates that ADP release largely depends on the forcelevel. (Note that this factor becomes predominant at the force level

higher than ∼ 1 pN in 1 mM ATP or ∼ 2 pN in10 µM ATP, as observed in Fig. 3f.)

Frequency of the intermediate stateThe occurrence frequency of the intermediatestate within the 36-nm main step is greatest inthe presence of BDM and is increased withforce irrespective of the nucleotide condition(Fig. 3g). In the absence of BDM, the propor-tion merged at a high force level irrespectiveof the nucleotide conditions, whereas at a lowforce level the proportion was decreased onlowering the ATP concentration and by theaddition of ADP.

Effects of BDM on the ATPase kinetics of myosin VActin filaments activate the steady-state ATPase activity of MV-1IQ(Fig. 4a). The solid line in Fig. 4a is the best fit to a hyperbola (rate = (Vmax × [actin] / (KATPase + [actin])). In the absence of BDM,the maximum turnover rate, Vmax, was 13.7 ± 0.9 s–1 and the KATPasewas 3.5 ± 0.6 µM, in agreement with earlier determinations12,21,22.BDM (100 mM) reduces the Vmax and KATPase approximately two-foldto 6.9 ± 0.2 s–1 and 2.3 ± 0.2 µM, respectively. In agreement with an earlier study23, ∼ 10 mM BDM has no appreciable effect on the maximal steady-state cycling of MV-1IQ.

Actin filaments accelerate the rate of transient Pi release from MV-1IQ-ADP-Pi (Fig. 4b). Time courses of Pi release follow singleexponentials (Fig. 4b inset) because myosin V is limited to a singleATP turnover by including excess ADP in the actin filament solution(see Methods). In the absence of BDM, the maximum rate of Pi releasefrom MV-1IQ-ADP-Pi (k+4′ following nomenclature of ref. 24) was 60 ± 9 s–1 and the actin concentration at the half-maximum rate (K9

–1, ref. 24) was 7.3 ± 0.6 µM. The rate of myosin V-ADP-Pi binding

880 VOLUME 11 NUMBER 9 SEPTEMBER 2004 NATURE STRUCTURAL & MOLECULAR BIOLOGY

Table 1 Parameters obtained by dwell time analysis

τTotal = τD exp(FdD / kBT) + τC + τi exp(Fdi / kBT)

Nucleotide states τTotal (F = 0) τD (ms), dD (nm) τC (ms) τi(ms), di (nm)

1 mM ATP 86.7 ± 13.0 1.1 ± 0.3, 12.5 ± 0.4 79.7 ± 11.5 5.9 ± 1.2, 2.2 ± 0.2

1 mM ATP + 100 mM BDM 141.2 ± 16.2 1.7 ± 0.4, 12.5 ± 0.4 110.6 ± 9.2 28.9 ± 6.6, 2.2 ± 0.2

1 mM ATP + 200 µM ADP 145.5 ± 20.2 2.8 ± 0.3, 12.5 ± 0.4 136.8 ± 18.7 5.9 ± 1.2, 2.2 ± 0.2

10 µM ATP 553.5 ± 50.1 1.1 ± 0.3, 12.5 ± 0.4 546.5 ± 48.6 5.9 ± 1.2, 2.2 ± 0.2

The main dwell time was fit to the sum of a force-dependent exponential and a constant, whereas the intermediatedwell time was fit to a force-dependent single exponential. The values of parameters τD, dD, τC, τi and di weredetermined by global fit with nonlinear optimization19,20 using SigmaPlot 8.0. For the global fit analysis, we assumedthat the values of dD and di are common to the main and intermediate states, respectively, and τD and τi for somenucleotide states.

Figure 4 Effects of BDM on ATPase cycle kinetics of single-headed myosinV. (a) Actin concentration dependence of MV-1IQ steady-state ATPaseactivity. The final [MV-1IQ] was 13 nM. The solid line is the best fit to ahyperbola. Error bars represent standard errors in the best fits to the steady-state ATPase activity. (b) Actin concentration dependence of the rate oftransient Pi release from MV-1IQ. The solid line is the best fit of the data to a hyperbola. Error bars represent standard errors in the best fits of the timecourses of Pi release to single exponentials. Inset, time course of Pi releasefrom 0.5 µM MV-1IQ-ADP-Pi after mixing with 5 µM F-actin and 2 mM ADP.The trace represents the raw, unaveraged data. The solid line through thedata is the best fit to a single exponential with a rate of 28.5 ± 0.6 s–1.(c) Time courses of mantADP release from 0.25 µM actomyosin V-1IQ withand without BDM. The jagged lines are the averages of three traces. Thesmooth lines through the data are the best fits to single exponentials.

©20

04 N

atur

e P

ublis

hing

Gro

up

http

://w

ww

.nat

ure.

com

/nat

stru

ctm

olbi

ol

A R T I C L E S

from D* to D (a part of the 24-nm substep may be attributable to thebinding of detached head (T or DPi) to actin after Brownian motion).It is notable that one set of researchers30 showed that, in a single-headed myosin V, there exist two stable binding modes with differentangles of the neck region in the ADP-bound state. Also, anothergroup31 recently showed that a one-headed myosin V produces a 25-nm working stroke.

The occurrence frequency of the 36-nm main step, in which theintermediate state could not be identified, was substantially larger thanthat predicted from the frequency extrapolated from the exponentialfitting of the distribution of the dwell time of the intermediate states(Fig. 3a,b). This is understandable if we assume that there are twokinetic pathways for the 36-nm steps with (pathway 1) and without(pathway 2) passing through the intermediate state.

On the main state, the dwell time is largely prolonged on loweringthe ATP concentration, and by adding ADP or 100 mM BDM(Fig. 3d–f), implying that it is shortened by the attachment of ATP andthe detachment of ADP. The simplest assumption deduced from theseresults is that the main state is terminated; that is, the 12-nm substepoccurs upon the ATP binding to a bound trailing head of myosin Vfrom which ADP has been released. This assumption is similar to thatpredicted by Kolomeisky and Fisher32 and also consistent with severalmodels previously proposed for myosin V7,22,24,26,27 and for myosinVI24. This is also consistent with the effect of BDM that slows the ADPrelease (Fig. 4c) and the overall ATPase activity as shown by thedecrease in the velocity of myosin V movement (data not shown). Thatis, BDM stabilizes not only the D* state but also the D state.

Thus, we propose the hand-over-hand model as illustrated in Fig. 5,which incorporates the previous models7,26,27 and which can explain, atleast qualitatively, all the data presented here. First, the model canaccount for the existence of two kinds of steps: the 36-nm step com-posed of the sequential 12-nm and 24-nm substeps (pathway 1), andthe 36-nm main step without substeps (pathway 2). The occurrencefrequency of these two pathways depends on the conditions as dis-cussed later. According to this model, the intermediate state is in a single-headed binding, whereas the main state is in a double-headedbinding. This may be experimentally confirmed by the measurementsof the stiffness of the protein-bead complex. In this respect, it is notable

NATURE STRUCTURAL & MOLECULAR BIOLOGY VOLUME 11 NUMBER 9 SEPTEMBER 2004 881

to actin filaments (K9k+4′) was ∼ 8 µM–1 s–1, comparable to earliermeasurements12. In the presence of 100 mM BDM, k+4′ was 59 ± 1.4 s–1 and K9

–1 was 7.7 ± 0.6 µM. BDM (100 mM) does not affectk+4′ or K9

–1 (Fig. 4b).ADP release limits steady-state cycling of myosin V12,21. The rate of

N-methylanthraniloyl-ADP (mantADP) release from actoMV-1IQ inthe absence of BDM (Fig. 4c) was 10.2 ± 0.1 s–1, in agreement with earlier determinations12,21,22. BDM (100 mM) slowed the rate ofmantADP release approximately two-fold to 4.6 ± 0.1 s–1. The reduc-tion in ADP release accounts for the slower turnover rate in the pres-ence of 100 mM BDM.

The KATPase of a myosin with rate-limiting ADP release (k+5′ = Vmax)can be related to the maximum rate of Pi release (k+4′), the actin con-centration needed to reach half of the maximum Pi release rate (K9, inunits of M–1), and the equilibrium constant for ATP hydrolysis(K3)(ref. 21): KATPase = k+5′(K9/k+4′) ({1 + K3}/K3). The two-foldreduction of ADP release (k+5′) by 100 mM BDM accounts for thetwo-fold reduction in Vmax and KATPase, suggesting that BDM does notgreatly affect the equilibrium constant for ATP hydrolysis (K3), eventhough it was not measured directly. This result is in contrast to thecase of muscle myosin II, where BDM stabilizes the actomyosin ADP-Pi state and accordingly suppresses Pi release25. Finally, we con-firmed that the velocity of bead movement in the absence of externalload decreased as the concentrations of BDM increased in the presenceof 1 mM ATP. The BDM concentration dependence of the averagebead velocity showed that the velocity decreases to nearly half with theaddition of 100 mM BDM (data not shown). These results are consis-tent with those of the steady-state ATPase (Fig. 4a).

DISCUSSIONWe propose here a hand-over-hand walking model of myosin V by taking into account all the experimental data described here togetherwith those reported elsewhere7,26,27.

First, considering that BDM reduces the rate of ADP release fromactomyosin V (Fig. 4c) and largely prolongs the intermediate state(Fig. 3a–c), it is possible that the main species of the intermediate stateis the actomyosin V–ADP complex and that the 24-nm substeps occuraccompanied by the ADP release. However, this contradicts the obser-vation that exogenous ADP does not affect the dwell time of the inter-mediate state. To resolve this apparent contradiction, we assume thatisomerization exists in the actomyosin V (AM)–ADP complex, that is,AM*ADP (D*) and AMADP (D). It has already been proposed on thebasis of a kinetic study using mant-nucleotides13 that two states existin the AM-ADP complex. Furthermore, it is well known that such anisomerization exists in the contraction mechanism of the muscle acto-myosin II system28. Here, D is a complex produced by the binding ofADP to AM, whereas D* is produced after the hydrolysis of ATP on theAM-ATP complex28,29. We assume that D* is the major component ofthe intermediate state, implying that BDM stabilizes the D* state.

The intermediate state is terminated by the 24-nm substep, so that astraightforward deduction from such consideration is that the 24-nmsubstep occurs by the mechanical step mainly following the transition

Figure 5 Hand-over-hand model coupled with nucleotide states explainingthe present results. We propose that there are two possible pathways:pathway 1 in which 12-nm and 24-nm substeps occur (left column), andpathway 2 in which only the 36-nm main step occurs (right column). BDM is assumed to stabilize both the AMD complex (D) and the AM*D complex(D*), so that the transition rates indicated by red arrows are slowed down. T, ATP; D, ADP; Pi, inorganic phosphate; φ,no nucleotides; (Pi), a possiblestep at which Pi release occurs. For more details, see the text.

©20

04 N

atur

e P

ublis

hing

Gro

up

http

://w

ww

.nat

ure.

com

/nat

stru

ctm

olbi

ol

A R T I C L E S

that Veigel et al.31 reported that the low stiffness intervals, which mayimply the single-headed binding, exist during the main steps.

Second, the 12-nm substep is assumed to occur upon the bindingof ATP to the bound trailing head. Although we do not know whetherthe bound leading head is D* or D*Pi at this stage (the upper left inFig. 5), we infer that the 12-nm step may be attributable to the con-formational change of the leading head due to the transition from DPito D* (or D*Pi).

Third, if the transition rate from D* to D decreases on increasing theforce level, not only the dwell time of the intermediate state but alsothat of the main state are prolonged. Besides, we can understand thatthe higher the force level, the higher the occurrence frequency of theintermediate state irrespective of the nucleotide conditions (Fig. 3g),because the transition probability of the process shown by the hori-zontal arrows in Fig. 5 (from the pathway 1 to the pathway 2) is sup-pressed irrespective of the nucleotide conditions.

Fourth, because we assumed that BDM stabilizes not only the D* state but also the D state, both the intermediate state and the mainstate are stabilized, so that the dwell times of both states are expected tobe prolonged. This assumption can also explain a large degree ofextension of the dwell time, τi, of the intermediate state, because it istightly coupled to the lifetime of the D* state, whereas the degree ofextension of the dwell time of the main state was less because severalother states must be involved in the main state.

Fifth, at low ATP concentrations, the attachment of ATP to thebound trailing head is slowed down, so that the probability increasesthat the transition from D* to D occurs at the bound leading headbefore ATP binds to the trailing head. This results in the increase in theoccurrence frequency of the pathway 2. In the pathway 2, it is expectedthat the internal strain is largest within the D-D complex, because theleading head is considered to take the orientation similar to that real-ized after the 24-nm mechanical step. Thus, the D-D (also φ-D) com-plex may have a telemark shape as observed by electron microscopy7.It is to be stressed that the two pathways are not independent of eachother, but the pathway 1 (or 2) is chosen when the transition from D* to D occurs after (or before) the 12-nm substep; that is, the modelin Fig. 5 proposes that the timing of the 12-nm substep and the iso-merization (the transition from D* to D) determines the pathways.

Finally, if the binding affinity of ADP for the bound trailing head islower than that for the bound leading head because of themechanochemical coupling due to the internal strain, ADP tends todetach from the bound trailing head, which results in the binding ofATP to the trailing head. Such an asymmetrical binding affinity ofADP may be prerequisite for the directional movement of myosin Vtoward the barbed end of an actin filament, although this must beexperimentally confirmed. We infer that the loading-direction depen-dency of binding affinity of ADP assumed for myosin V is the reverseof that for kinesin as recently reported by us20. Also, another study33

showed that the binding of ATP to the bound leading head is pre-requisite for the directional movement of kinesin toward the plus-endof a microtubule. The correspondence between internal strain andnucleotide affinity in mechanochemical coupling for myosin V may bedifferent from that for kinesin.

METHODSProtein and assays used for optical trapping measurements. Myosin V waspurified from chick brains34. Actin purified from rabbit skeletal muscle wasbiotinylated by 10% and, after polymerization, the filaments were labeled withrhodamine phalloidin (Molecular Probes)35. About 1 nM of fluorescent poly-styrene beads (200 nm in diameter, yellow-green; Molecular Probes) were incubated for 20 min in an assay buffer (10 mM imidazole-HCl, pH 7.2, 75 mM

KCl, 2.5 mM MgCl2, 2 mM DTT and 0.1 mM EGTA) containing 10 mg ml–1

BSA. Myosin V (650 kDa) molecules were mixed with the beads at a molar ratioof 3:1 in assay buffer containing 300 mM KCl instead of 75 mM KCl as had beenused in the previous studies. The average number of functional myosin V mol-ecules on a bead was estimated by statistical analyses to be one (considering thegeometry of the myosin V on the bead, we estimate that only single myosin Vmolecules interacted with an actin filament in almost all the measure-ments17,20). Assay buffer containing biotinylated BSA at 3 mg ml–1 was intro-duced into a flow cell and incubated for 2 min to coat the glass surface withbiotinylated BSA. After rinsing with two volumes of assay buffer, 2 mg ml–1

streptavidin in assay buffer was flowed into the cell and incubated for 2 min.After rinsing with assay buffer, a solution of actin filaments, of which 10% wasbiotinylated and labeled with rhodamine-phalloidin, was flowed into the cell toallow binding of the filaments to the glass surface through avidin-boundbiotinylated BSA. The flow cell was then filled with assay solution containingthe myosin V–coated beads, filtered BSA and an oxygen-scavenging enzymesystem9. The final solvent condition was approximately 0.1 pM myosinV–coated beads, 10 mM imidazole-HCl, pH 7.2, 75 mM KCl, 2.5 mM MgCl2,2 mM DTT, 0.1 mM EGTA, 3.6 mg ml–1 glucose, 0.08 mg ml–1 glucose oxidase,0.01 mg ml–1 catalase, 0.95% (v/v) β-mercaptoethanol and nucleotides (1 mMATP, 10 µM ATP, 1 mM ATP + 200 µM ADP or 1 mM ATP + 100 mM BDM).We were able to observe repeatedly the stepwise movements of myosinV–coated beads along the same actin filaments on the same beads, presumablyfor the same myosin V molecules, by using optical tweezers to manipulate thebeads. All experiments on microscopy were done at 24 ± 1 °C.

We found that the percentage of biotinylation of actin is important for theprocessive movement of myosin V. In other words, myosin V could move pro-cessively on 1% and 10% biotinylated actin filaments, whereas it could not on100% biotinylated ones, suggesting that the manner in which actin filamentsbind to the glass surface is crucial. It should be noted that myosin V is reportedto be a left-handed spiral motor9, so that the revolution of a myosin V–coatedbead around the actin filament could be sterically hindered. However, this pos-sibility could be ignored, because the maximum distance of bead displacementwas less than ∼ 0.3 µm (Fig. 1b), such that it was much shorter than the dis-tance, 2 µm, for one revolution9.

The velocity of myosin V under no external load was obtained from the timecourse of bead movement along an actin filament in the absence of optical trap.The bead position was determined by the center of the fluorescence intensitydistribution of the bead every video frame.

Proteins and reagents used for biochemical experiments. Actin was purifiedfrom rabbit skeletal muscle and gel filtered over Sephacryl S-300HR (ref. 12).The motor domain of myosin V containing the first IQ motif (MV-1IQ) andthe essential light chain, LC-1sa, were co-expressed in Sf9 insect cells and puri-fied by FLAG affinity chromatography12. The fluorescently labeled mutant ofthe phosphate-binding protein (MDCC-PBP; clone provided by M.R. Webb,National Institute for Medical Research, London) was expressed, purified andlabeled as described36. ATP and ADP were purchased from Roche MolecularBiochemicals. mantADP was synthesized as described37 or purchased fromMolecular Probes with identical results. A molar equivalent of MgCl2 wasadded to nucleotides immediately before use. BDM was purchased from Sigma(lot 092K1722) and prepared as a 250 mM stock solution in KMg50-MOPS (10 mM MOPS, pH 7.0, 50 mM KCl, 1 mM MgCl2, 1 mM EGTA and 1 mMDTT) immediately before use21.

Instrumentation and calibration. The myosin V–coated bead was trapped withan optical tweezers—that is, a focused infrared laser beam (λ = 1,064 nm, 2 W;Spectra Physics)—and illuminated diagonally by a focused red laser beam (λ = 685 nm, 20 mW; Phototechnica) through an objective lens (fluor ×100/1.3oil; Nikon). The light scattered by the bead was gathered by an objective lenswith an aperture (NA = 0.5, ×100 oil; Olympus) and projected onto a quadrantphotodiode sensor (S4349; Hamamatsu Photonics) coupled to a differentialamplifier (20-kHz roll-off frequency; OP711, Sentec). The fluorescently labeledbeads and actin labeled with rhodamine-phalloidin were excited by a greenlaser (λ = 532 nm, 50 mW; Peace Engineering), and the fluorescence imageswere captured by a silicon-intensified target camera (C-2740; HamamatsuPhotonics) and displayed on a video monitor. The bead positions were

882 VOLUME 11 NUMBER 9 SEPTEMBER 2004 NATURE STRUCTURAL & MOLECULAR BIOLOGY

©20

04 N

atur

e P

ublis

hing

Gro

up

http

://w

ww

.nat

ure.

com

/nat

stru

ctm

olbi

ol

A R T I C L E S

recorded on a computer equipped with a laboratory interface board (MacLab;AD Instruments)15 at a sampling rate of 10 kHz through a digital low-pass filterat 10 kHz. The bead displacement was calibrated by moving the photodiode15.The trap stiffness of the optical tweezers was calibrated for every bead from the standard deviation of the position fluctuation of the trapped bead(0.009–0.011 pN nm–1). The step size was obtained directly from individualstepwise movements of the bead at a sampling time of 0.1 ms and estimated asthe difference between the average bead positions determined for 5 ms eachinterval just before and after the steps. We did not take into account the attenu-ation factor, which is a function of the stiffness of optical trap and the stiffnessof the protein-bead complex15,17, because the trap stiffness we used is consid-ered to be much smaller than that of the protein-bead complex. The averagevelocity of bead movement (v) was estimated by dividing the average step size(36 nm) by the average total dwell time (τTotal in units of ms) at each externalload, which is balanced to the force (F) generated by myosin V. Force-velocityrelationships were described by the following function: v = 36 nm / τTotal =36/{τD exp(FdD / kBT)+ τC +τi exp(Fdi / kBT)} (nm ms–1), where τD, dD, τC, τi,di, kB and T are described in the text.

Steady-state and transient kinetic experiments. All kinetic experiments weredone at 25 ± 0.1 ºC in KMg50-MOPS with an SX.18MV-R stopped-flow appa-ratus (Applied Photophysics). Fitting was done with Pro-K software providedwith the instrument. Steady-state ATPase activity of MV-1IQ was measuredusing the ATP-regenerating, NADH-coupled assay as described21. BDM hadminimal effects on the assay components as determined by direct mixing withMgADP.

Transient Pi release of MV-1IQ was measured using MDCC-PBP (λex = 430 nm, 455 nm emission filter) with the instrument in sequential mix-ing mode as described24. Briefly, 2 µM MV-1IQ (treated with 0.01 U ml–1

potato grade VII apyrase to remove residual ADP and ATP, ± 200 mM BDM)was mixed with 300 µM MgATP (± 200 mM BDM) and aged for 40–60 ms toallow for nucleotide binding and hydrolysis to occur, then mixed with an equalvolume of a range of actin filament concentrations. As first described formyosin VI24, myosin V was limited to a single ATP turnover by including 2 mMMgADP with the actin. ADP competes with ATP for binding to myosin V afterthe first turnover and inhibits subsequent steady-state cycling. Therefore, timecourses follow single exponentials (see Fig. 4b inset) rather than exponentialsfollowed by a linear steady-state component, permitting more accurate fitting of the Pi release time courses. This method of measuring Pi release can beused for all high-duty-ratio myosins with rapid rates of ADP binding and highADP affinities.

ADP release was measured with a fluorescent nucleotide mantADP12. Thefluorescence of mantADP (λex = 366 nm, 400-nm emission filter) was moni-tored after an equilibrated mixture of 0.5 µM actomyosin V-1IQ (± 200 mMBDM) and 20 µM mantADP was mixed with an equal volume of 2 mMMgADP.

ACKNOWLEDGMENTSWe thank M.R. Webb for the phosphate-binding protein clone and suggestions on purification and labeling, and H.L. Sweeney for providing the myosin V heavyand light chain viruses. We are grateful to N. Sasaki, M.Y. Ali, K. Kinosita, Jr. and E.M. Ostap for encouragement and stimulating discussions. This research waspartly supported by Grants-in-Aid for Specially Promoted Research, for the Bio-venture Project and for The 21st Century COE Program (Physics of Self-Organization Systems) at Waseda Univ. from the Ministry of Education, Sports,Culture, Science and Technology of Japan (to S.I.) and supported by a ScientistDevelopment Grant from the American Heart Association and a grant from the USNational Science Foundation (to E.M.D.L.C.). S.U. is a postdoctoral fellow of theJapan Society for the Promotion of Science. A.O.O. is supported by a Cellular &Molecular Biology graduate training grant (Yale University).

COMPETING INTERESTS STATEMENTThe authors declare that they have no competing financial interests.

Received 31 March; accepted 18 June 2004Published online at http://www.nature.com/nsmb/

1 Cheney, R.E. et al. Brain myosin V is a two-headed unconventional myosin with motoractivity. Cell 75, 13–23 (1993).

2. Miller, K.E. & Sheetz, M.P. Characterization of myosin V binding to brain vesicles.J. Biol. Chem. 275, 2598–2606 (2000).

3. Vale, R.D. The molecular motor toolbox for intracellular transport. Cell 112, 467–480(2003).

4. Reck-Peterson, S., Provance, D.W., Mooseker, M.S. & Mercer, J.A. Review: class Vmyosins. Biochim. Biophys. Acta 1496, 36–51 (2000).

5. Sakamoto, T., Amitani, I., Yokota, E. & Ando, T. Direct observation of processivemovement by individual myosin V molecules. Biochem. Biophys. Res. Commun. 272,586–590 (2000).

6. Mehta, A.D. et al. Myosin V is a processive actin-based motor. Nature 400, 590–593(1999).

7. Walker, M. et al. Two-headed bindings of a processive myosin to F-actin. Nature 405,804–807 (2000).

8. Rief, M. et al. Myosin V stepping kinetics: a molecular model for processivity. Proc.Natl. Acad. Sci. USA 97, 9482–9486 (2002).

9. Ali, M.Y. et al. Myosin V is a left-handed spiral motor on the right-handed actin helix.Nat. Struct. Biol. 9, 464–467 (2002).

10. Forkey, J.N., Quinlan, M.E., Shaw, M.A., Corrie, J.E. & Goldman, Y.E. Three-dimen-sional structural dynamics of myosin V by single-molecule fluorescence polarization.Nature 422, 399–404 (2003).

11. Yildiz, A. et al. Myosin V walks hand-over-hand: single fluorophore imaging with 1.5-nm localization. Science 300, 2061–2065 (2003).

12. De La Cruz, E.M., Wells, A.L., Rosenfeld, S.S., Ostap, E.M. & Sweeney, H.L. Thekinetic mechanism of myosin V. Proc. Natl. Acad. Sci. USA 96, 13726–13731(1999).

13. Trybus, K.M., Krementsova, E. & Freyzon, Y. Kinetic characterization of a monomericunconventional myosin V construct. J. Biol. Chem. 274, 27448–27456 (1999).

14. Higuchi, H. & Takemori, S. Butanedione monoxime suppresses contraction andATPase activity of rabbit skeletal muscle. J. Biochem. 105, 638–643 (1989).

15. Nishiyama, M., Higuchi, H. & Yanagida, T. Chemomechanical coupling of the forwardand backward steps of single kinesin molecules. Nat. Cell Biol. 4, 790–797 (2002).

16. Moore, J.R., Krementsova, E.B., Trybus, K.M. & Warshaw, D.M. Myosin V exhibits ahigh duty cycle and large unitary displacement. J. Cell Biol. 155, 625–635 (2001).

17. Kojima, H., Muto, E., Higuchi, H. & Yanagida, T. Mechanics of single kinesin mole-cules measured by optical trapping nanometry. Biophys. J. 73, 2012–2022 (1997).

18. Svoboda, K., Schmidt, C.F., Schnapp, B.J. & Block, S.M. Direct observation ofkinesin stepping by optical trapping interferometry. Nature 365, 721–727 (1993).

19. Schnitzer, M.J., Visscher, K. & Block, S.M. Force production by single kinesin motors.Nat. Cell Biol. 2, 718–723 (2000).

20. Uemura, S. & Ishiwata, S. Loading direction regulates the affinity of ADP for kinesin.Nat. Struct. Biol. 10, 308–311 (2003).

21. De La Cruz, E.M., Sweeney, H.L. & Ostap, E.M. ADP inhibition of myosin V ATPaseactivity. Biophys. J. 79, 1524–1529 (2000).

22. De La Cruz, E.M., Wells, A.L., Sweeney, H.L. & Ostap, E.M. Actin and light chain iso-form dependence of myosin V kinetics. Biochemistry 39, 14196–14202 (2000).

23. Ostap, E.M. 2,3-Butanedione monoxime (BDM) as a myosin inhibitor. J. Muscle Res.Cell Motil. 23, 305–308 (2002).

24. De La Cruz, E.M., Ostap, E.M. & Sweeney, H.L. Kinetic mechanism and regulation ofmyosin VI. J. Biol. Chem. 276, 32373–32381 (2001).

25. Herrmann, C., Wray, J., Travers, F. & Barman, T. Effect of 2,3-butanedione monoximeon myosin and myofibrillar ATPases. An example of an uncompetitive inhibitor.Biochemistry 31, 12227–12232 (1992).

26. Spudich, J.A. & Rock, R.S. A crossbridge too far. Nat. Cell Biol. 4, E8–E10 (2002).27. Vale, R.D. Myosin V motor proteins: marching stepwise towards a mechanism. J. Cell

Biol. 163, 445–450 (2003).28. Goldman, Y.E. and Brenner, B. Special topic: molecular mechanism of muscle con-

traction. Annu. Rev. Physiol. 49, 629–636 (1987).29. Dantzig, J.A., Hibberd, M.G., Trentham, D.R. & Goldman, Y.E. Cross-bridge kinetics

in the presence of MgADP investigated by photolysis of caged ATP in rabbit psoasmuscle fibres. J. Physiol. 432, 639–680 (1991).

30. Burgess, S. et al. The prepower stroke conformation of myosin V. J. Cell Biol. 159,983–991 (2002).

31. Veigel, C., Wang, F., Bartoo, M.L., Sellers, J.R. & Molloy, J.E. The gated gait of theprocessive molecular motor myosin V. Nat. Cell Biol. 4, 59–65 (2002).

32. Kolomeisky, A.B. & Fisher, M.E. A simple model describes the processivity of MyosinV. Biophys. J. 84, 1642–1650 (2003).

33. Rosenfeld, S.S., Fordyce, P.M., Jefferson, G.M., King, P.H. & Block, S.M. Steppingand stretching. How kinesin uses internal strain to walk processively. J. Biol. Chem.278, 18550–18556 (2003).

34. Cheney, R.E. Purification and assay of myosin V. Methods Enzymol. 293, 3–18(1998).

35. Yanagida, T., Nakase, M., Nishiyama, K. & Oosawa, F. Direct observation of motion ofsingle F-actin filaments in the presence of myosin. Nature 307, 58–60 (1984).

36. Brune, M. et al. Mechanism of inorganic phosphate interaction with phosphate bind-ing protein from Escherichia coli. Biochemistry 37, 10370–10380 (1998).

37. Hiratsuka, T. New ribose-modified fluorescent analogs of adenine and guaninenucleotides available as substrates for various enzymes. Biochim. Biophys. Acta 742,496–508 (1983).

NATURE STRUCTURAL & MOLECULAR BIOLOGY VOLUME 11 NUMBER 9 SEPTEMBER 2004 883

©20

04 N

atur

e P

ublis

hing

Gro

up

http

://w

ww

.nat

ure.

com

/nat

stru

ctm

olbi

ol